This may be a trivial question, but...
Often when comparing analysis of samples by a selective methodology (e.g. GC-NPD) with GC-MS, the latter appears to give an enhanced response. In both cases, the samples are quantified against internal standards which are not matrix matched. Can anybody give any insight into what may cause this signal enhancement observed in the MS detection?

I am assuming that you are doing SIM analysis, since scan mode in MS is usually significantly less sensitive than a selective detector. SIM mode is generally about as sensitive as an NPD, etc., if the threshold is set properly. If the threshold is set to high, the noise is not seen and the signal to noise looks better than it really is.

You are very likely seeing injector/column enhancement effects, unless you are using identical columns--usually not the case. The enhancement effect is more dramatic on the small bore columns usually used in MS.

Run exactly the same series of injections on both systems, with standards before and after the samples, and compare the results.

There is another post, 'Abnormal response on standard and samples August 1 - 11:13 am' dealing with what may be the same issue. It is a common problem with trace analysis of biological materials.

NPDs and SIM-MS are 2 entirely different detection technologies. If I understand your question properly, SIM-MS may simply have better detection limits! If the analogous question were asked about FIDs and TCDs, I would say the same thing. I'll use the FID/TCD comparison for illustration: Basically, the physical process of ionizing (burning) molecules, usually >1 ion/molecule, ion collection efficiencies, and subsequent current measurement is apparently more efficient than measuring resistance changes due to environmental heat capacity changes and the effect on the TCD's wire. Apples and oranges, but each detector has its merits and limitations.

Do watch out for what Anonymous number one said though when looking at S/N though. If you ever see a truly flat baseline when you zoom-in on the baseline, something is either wrong (A/D problems or incorrect digital gain settings) or, if you are developing a new type of detector, write it up and notify the Nobel-prize selection committee, because you just achieved what was formerly thought to be impossible according to the laws of physics.